We think now is the time to reduce exposure to NAND and
Asian semiconductor names as the industry has benefitted
from sizeable demand tailwinds and unprecedented pricing
power, which we see reversing soon. We downgrade Samsung,
TSMC and WDC to EW; initiate Phison at UW.
We believe the NAND cycle is entering a downturn for the first time since 1Q16
while the DRAM cycle should remain strong through 1Q18. Our channel checks
last week suggest that NAND pricing is likely to fall earlier than the market
expects, with early signs of demand destruction from mobile content and
accelerating supply overwhelming enterprise SSD demand. This translates to
poor risk-reward for NAND stocks, and with valuation restrained by pricing risk,
we make several downgrades.
Downgrade Samsung, TSMC and WDC to EW; initiate Phison at UW: Given our
view of the cycle, we cannot recommend the sector until the market recognizes
mounting pressure on NAND prices and slowing logic chip growth momentum in
the near term. The long-term NAND story remains attractive and stocks can
manage through a period of reduced prices and returns, but both factors argue
against a premium rating. Our quantitative analysis of prior memory cycles
suggests that the best time to reduce exposure to stocks is 3-6 months ahead of
the peak of DRAM prices, as market skepticism hinders share-price performance
closer to the turning point. This will play out in the next 1-2 quarters, we think.
We expect DRAM to remain strong in 2018... We maintain our bullish view on
DRAM as our short-term checks stay robust, especially for server, and would add
to Micron positions on any NAND-related weakness. Unlike NAND, we see strong
conditions persisting through 1H18 for DRAM as capital spending has been strong
but not excessive and demand should remain healthy, with large content
increases in servers and phones offsetting modest downward pressure from PCs.
Moving into 2019-20, we expect a period of oversupply as capital spending will
ramp to excessive levels during 2018, leading to significant increases in supply
towards the end of the year.
...but a lot may be priced in and there are potential headwinds: We often hear
that 'DRAM is fundamentally different because of THE CLOUD'. Simply put, the
market is effectively taking the opposite stance from the December 2015 trough,
when extreme bearishness prevailed but being overweight the sector would have
paid huge dividends. Using the same argument, one should do the opposite by
selling the rally now that DRAM stocks are up 182% since then. While the market
firmly believes history will not repeat, we identify risks that could impact the
87% of the DRAM market that is non-cloud, including: (1) demand pull-in from
iPhone X and Chinese smartphone content changes, (2) supply growth impact
from technology migration challenges, and (3) Samsung's pace of supply growth.